9. Fair synthetic data and ethical algorithms: the fairness conversation with Paul Tiwald, Head of Data Science at MOSTLY AI


Manage episode 293972165 series 2895967
Player FM과 저희 커뮤니티의 MOSTLY AI 콘텐츠는 모두 원 저작자에게 속하며 Player FM이 아닌 작가가 저작권을 갖습니다. 오디오는 해당 서버에서 직접 스트리밍 됩니다. 구독 버튼을 눌러 Player FM에서 업데이트 현황을 확인하세요. 혹은 다른 팟캐스트 앱에서 URL을 불러오세요.

Paul Tiwald has been part of the MOSTLY AI team since the beginning. He is the mastermind behind the team's research into fairness and the idea of fair synthetic data.

In this episode, you will hear about:

  • what it's like to work in the field of artificial intelligence (spoiler: it's really fun!)
  • how the idea of fair synthetic data came up
  • how to create machine learning models that are private and fair by design
  • why is it so challenging to remove bias from an algorithm
  • what are proxy variables, and why are they dangerous
  • what is the definition of fairness, and why do we need one in the first place
  • how should companies start implementing fairness and ethical approaches into their AI development
  • why it's impossible to fix bias without fair synthetic data and algorithmic fairness

12 에피소드